Environment Independent Speech Recognition System

Abstract—- Speech recognition is a method of finding similarity between two sequences. Various researches have been done on it. In our research, we are trying to achieve the optimal accuracy during the recognition procedure. Here, we are extracting features of the voice sample before filtering it through a noise reduction filter. For each individual, there are number of features are taken using feature extraction algorithm called mel frequency cepstral co-efficient (MFCC). After extracting the features of all the training samples, we have taken the average. Now extracting the features of the testing sample, similarities are calculated using dynamic time warping. We are comparing the results of frame error, word error of the previous work with our research